We present in this paper one of the largest galaxy morphological classification catalogues to date, including over 20 million of galaxies, using the Dark Energy Survey (DES) Year 3 data based on ...Convolutional Neural Networks (CNN). Monochromatic \(i\)-band DES images with linear, logarithmic, and gradient scales, matched with debiased visual classifications from the Galaxy Zoo 1 (GZ1) catalogue, are used to train our CNN models. With a training set including bright galaxies (\(16\le{i}<18\)) at low redshift (\(z<0.25\)), we furthermore investigate the limit of the accuracy of our predictions applied to galaxies at fainter magnitude and at higher redshifts. Our final catalogue covers magnitudes \(16\le{i}<21\), and redshifts \(z<1.0\), and provides predicted probabilities to two galaxy types -- Ellipticals and Spirals (disk galaxies). Our CNN classifications reveal an accuracy of over 99\% for bright galaxies when comparing with the GZ1 classifications (\(i<18\)). For fainter galaxies, the visual classification carried out by three of the co-authors shows that the CNN classifier correctly categorises disky galaxies with rounder and blurred features, which humans often incorrectly visually classify as Ellipticals. As a part of the validation, we carry out one of the largest examination of non-parametric methods, including \(\sim\)100,000 galaxies with the same coverage of magnitude and redshift as the training set from our catalogue. We find that the Gini coefficient is the best single parameter discriminator between Ellipticals and Spirals for this data set.
In this work we present the galaxy clustering measurements of the two DES lens galaxy samples: a magnitude-limited sample optimized for the measurement of cosmological parameters, MagLim, and a ...sample of luminous red galaxies selected with the redMaGiC algorithm. MagLim / redMaGiC sample contains over 10 million / 2.5 million galaxies and is divided into six / five photometric redshift bins spanning the range \(z\in0.20,1.05\) / \(z\in0.15,0.90\). Both samples cover 4143 deg\(^2\) over which we perform our analysis blind, measuring the angular correlation function with a S/N \(\sim 63\) for both samples. In a companion paper (DES Collaboration et al. 2021)), these measurements of galaxy clustering are combined with the correlation functions of cosmic shear and galaxy-galaxy lensing of each sample to place cosmological constraints with a 3\(\times\)2pt analysis. We conduct a thorough study of the mitigation of systematic effects caused by the spatially varying survey properties and we correct the measurements to remove artificial clustering signals. We employ several decontamination methods with different configurations to ensure the robustness of our corrections and to determine the systematic uncertainty that needs to be considered for the final cosmology analyses. We validate our fiducial methodology using log-normal mocks, showing that our decontamination procedure induces biases no greater than \(0.5\sigma\) in the \((\Omega_m, b)\) plane, where \(b\) is galaxy bias. We demonstrate that failure to remove the artificial clustering would introduce strong biases up to \(\sim 7 \sigma\) in \(\Omega_m\) and of more than \(4 \sigma\) in galaxy bias.
The accuracy of photometric redshifts (photo-zs) particularly affects the results of the analyses of galaxy clustering with photometrically-selected galaxies (GCph) and weak lensing. In the next ...decade, space missions like Euclid will collect photometric measurements for millions of galaxies. These data should be complemented with upcoming ground-based observations to derive precise and accurate photo-zs. In this paper, we explore how the tomographic redshift binning and depth of ground-based observations will affect the cosmological constraints expected from Euclid. We focus on GCph and extend the study to include galaxy-galaxy lensing (GGL). We add a layer of complexity to the analysis by simulating several realistic photo-z distributions based on the Euclid Consortium Flagship simulation and using a machine learning photo-z algorithm. We use the Fisher matrix formalism and these galaxy samples to study the cosmological constraining power as a function of redshift binning, survey depth, and photo-z accuracy. We find that bins with equal width in redshift provide a higher Figure of Merit (FoM) than equipopulated bins and that increasing the number of redshift bins from 10 to 13 improves the FoM by 35% and 15% for GCph and its combination with GGL, respectively. For GCph, an increase of the survey depth provides a higher FoM. But the addition of faint galaxies beyond the limit of the spectroscopic training data decreases the FoM due to the spurious photo-zs. When combining both probes, the number density of the sample, which is set by the survey depth, is the main factor driving the variations in the FoM. We conclude that there is more information that can be extracted beyond the nominal 10 tomographic redshift bins of Euclid and that we should be cautious when adding faint galaxies into our sample, since they can degrade the cosmological constraints.
Analyses of type Ia supernovae (SNe Ia) have found puzzling correlations between their standardised luminosities and host galaxy properties: SNe Ia in high-mass, passive hosts appear brighter than ...those in lower-mass, star-forming hosts. We examine the host galaxies of SNe Ia in the Dark Energy Survey three-year spectroscopically-confirmed cosmological sample, obtaining photometry in a series of "local" apertures centred on the SN, and for the global host galaxy. We study the differences in these host galaxy properties, such as stellar mass and rest-frame \(U-R\) colours, and their correlations with SN Ia parameters including Hubble residuals. We find all Hubble residual steps to be \(>3\sigma\) in significance, both for splitting at the traditional environmental property sample median and for the step of maximum significance. For stellar mass, we find a maximal local step of \(0.098\pm0.018\) mag; \(\sim 0.03\) mag greater than the largest global stellar mass step in our sample (\(0.070 \pm 0.017\) mag). When splitting at the sample median, differences between local and global \(U-R\) steps are small, both \(\sim 0.08\) mag, but are more significant than the global stellar mass step (\(0.057\pm0.017\) mag). We split the data into sub-samples based on SN Ia light curve parameters: stretch (\(x_1\)) and colour (\(c\)), finding that redder objects (\(c > 0\)) have larger Hubble residual steps, for both stellar mass and \(U-R\), for both local and global measurements, of \(\sim0.14\) mag. Additionally, the bluer (star-forming) local environments host a more homogeneous SN Ia sample, with local \(U-R\) r.m.s. scatter as low as \(0.084 \pm 0.017\) mag for blue (\(c < 0\)) SNe Ia in locally blue \(U-R\) environments.
The apparent clustering in longitude of perihelion \(\varpi\) and ascending node \(\Omega\) of extreme trans-Neptunian objects (ETNOs) has been attributed to the gravitational effects of an unseen ...5-10 Earth-mass planet in the outer solar system. To investigate how selection bias may contribute to this clustering, we consider 14 ETNOs discovered by the Dark Energy Survey, the Outer Solar System Origins Survey, and the survey of Sheppard and Trujillo. Using each survey's published pointing history, depth, and TNO tracking selections, we calculate the joint probability that these objects are consistent with an underlying parent population with uniform distributions in \(\varpi\) and \(\Omega\). We find that the mean scaled longitude of perihelion and orbital poles of the detected ETNOs are consistent with a uniform population at a level between \(17\%\) and \(94\%\), and thus conclude that this sample provides no evidence for angular clustering.
The analysis of current and future cosmological surveys of type Ia supernovae (SNe Ia) at high-redshift depends on the accurate photometric classification of the SN events detected. Generating ...realistic simulations of photometric SN surveys constitutes an essential step for training and testing photometric classification algorithms, and for correcting biases introduced by selection effects and contamination arising from core collapse SNe in the photometric SN Ia samples. We use published SN time-series spectrophotometric templates, rates, luminosity functions and empirical relationships between SNe and their host galaxies to construct a framework for simulating photometric SN surveys. We present this framework in the context of the Dark Energy Survey (DES) 5-year photometric SN sample, comparing our simulations of DES with the observed DES transient populations. We demonstrate excellent agreement in many distributions, including Hubble residuals, between our simulations and data. We estimate the core collapse fraction expected in the DES SN sample after selection requirements are applied and before photometric classification. After testing different modelling choices and astrophysical assumptions underlying our simulation, we find that the predicted contamination varies from 5.8 to 9.3 per cent, with an average of 7.0 per cent and r.m.s. of 1.1 per cent. Our simulations are the first to reproduce the observed photometric SN and host galaxy properties in high-redshift surveys without fine-tuning the input parameters. The simulation methods presented here will be a critical component of the cosmology analysis of the DES photometric SN Ia sample: correcting for biases arising from contamination, and evaluating the associated systematic uncertainty.
This is the second in a series of three papers in which we present an end-to-end simulation from the MICE collaboration, the MICE Grand Challenge (MICE-GC) run. The N-body contains about 70 billion ...dark-matter particles in a \((3 \, h^{-1} \, {\rm Gpc})^3\) comoving volume spanning 5 orders of magnitude in dynamical range. Here we introduce the halo and galaxy catalogues built upon it, both in a wide (\(5000 \,{\rm deg}^2\)) and deep (\(z<1.4\)) light-cone and in several comoving snapshots. Halos were resolved down to few \(10^{11} \,h^{-1}\,{\rm M_{\odot}}\). This allowed us to model galaxies down to absolute magnitude M\(_r<-18.9\). We used a new hybrid Halo Occupation Distribution and Abundance Matching technique for galaxy assignment. The catalogue includes the Spectral Energy Distributions of all galaxies. We describe a variety of halo and galaxy clustering applications. We discuss how mass resolution effects can bias the large scale \(2\)-pt clustering amplitude of poorly resolved halos at the \(\lesssim 5\%\) level, and their \(3\)-pt correlation function. We find a characteristic scale dependent bias of \(\lesssim 6\%\) across the BAO feature for halos well above \(M_{\star}\sim 10^{12}\,h^{-1}\,{\rm M_{\odot}}\) and for LRG like galaxies. For halos well below \(M_{\star}\) the scale dependence at \(100\,{\rm Mpc} h^{-1}\) is \(\lesssim 2\%\). Lastly we discuss the validity of the large-scale Kaiser limit across redshift and departures from it towards nonlinear scales. We make the current version of the light-cone halo and galaxy catalogue (MICECATv1.0) publicly available through a dedicated web portal, http://cosmohub.pic.es, to help develop and exploit the new generation of astronomical surveys.
Forthcoming large photometric surveys for cosmology require precise and accurate photometric redshift (photo-z) measurements for the success of their main science objectives. However, to date, no ...method has been able to produce photo-\(z\)s at the required accuracy using only the broad-band photometry that those surveys will provide. An assessment of the strengths and weaknesses of current methods is a crucial step in the eventual development of an approach to meet this challenge. We report on the performance of 13 photometric redshift code single value redshift estimates and redshift probability distributions (PDZs) on a common set of data, focusing particularly on the 0.2--2.6 redshift range that the Euclid mission will probe. We design a challenge using emulated Euclid data drawn from three photometric surveys of the COSMOS field. The data are divided into two samples: one calibration sample for which photometry and redshifts are provided to the participants; and the validation sample, containing only the photometry, to ensure a blinded test of the methods. Participants were invited to provide a redshift single value estimate and a PDZ for each source in the validation sample, along with a rejection flag that indicates sources they consider unfit for use in cosmological analyses. The performance of each method is assessed through a set of informative metrics, using cross-matched spectroscopic and highly-accurate photometric redshifts as the ground truth. We show that the rejection criteria set by participants are efficient in removing strong outliers, sources for which the photo-z deviates by more than 0.15(1+z) from the spectroscopic-redshift (spec-z). We also show that, while all methods are able to provide reliable single value estimates, several machine-learning methods do not manage to produce useful PDZs. abridged
We demonstrate that highly accurate joint redshift-stellar mass probability distribution functions (PDFs) can be obtained using the Random Forest (RF) machine learning (ML) algorithm, even with few ...photometric bands available. As an example, we use the Dark Energy Survey (DES), combined with the COSMOS2015 catalogue for redshifts and stellar masses. We build two ML models: one containing deep photometry in the \(griz\) bands, and the second reflecting the photometric scatter present in the main DES survey, with carefully constructed representative training data in each case. We validate our joint PDFs for \(10,699\) test galaxies by utilizing the copula probability integral transform and the Kendall distribution function, and their univariate counterparts to validate the marginals. Benchmarked against a basic set-up of the template-fitting code BAGPIPES, our ML-based method outperforms template fitting on all of our predefined performance metrics. In addition to accuracy, the RF is extremely fast, able to compute joint PDFs for a million galaxies in just under \(6\) min with consumer computer hardware. Such speed enables PDFs to be derived in real time within analysis codes, solving potential storage issues. As part of this work we have developed GALPRO, a highly intuitive and efficient Python package to rapidly generate multivariate PDFs on-the-fly. GALPRO is documented and available for researchers to use in their cosmology and galaxy evolution studies.
In paper I of this series (Fosalba et al. 2013), we presented a new N-body lightcone simulation from the MICE collaboration, the MICE Grand Challenge (MICE-GC), containing about 70 billion ...dark-matter particles in a (3 Gpc)^3 comoving volume, from which we built halo and galaxy catalogues using a Halo Occupation Distribution and Halo Abundance Matching technique, as presented in the companion Paper II (Crocce et al. 2013). Given its large volume and fine mass resolution, the MICE-GC simulation also allows an accurate modeling of the lensing observables from upcoming wide and deep galaxy surveys. In the last paper of this series (Paper III), we describe the construction of all-sky lensing maps, following the "Onion Universe" approach (Fosalba et al. 2008), and discuss their properties in the lightcone up to z=1.4 with sub-arcmin spatial resolution. By comparing the convergence power spectrum in the MICE-GC to lower mass-resolution (i.e., particle mass ~ 10^11 Msun) simulations, we find that resolution effects are at the 5 % level for multipoles l ~ 10^3 and 20 % for l ~ 10^4. Resolution effects have a much lower impact on our simulation, as shown by comparing the MICE-GC to recent numerical fits by Takahashi et al 2012. We use the all-sky lensing maps to model galaxy lensing properties, such as the convergence, shear, and lensed magnitudes and positions, and validate them thoroughly using galaxy shear auto and cross-correlations in harmonic and configuration space. Our results show that the galaxy lensing mocks here presented can be used to accurately model lensing observables down to arcminute scales.